2.CNV calling

Here we use the R package epiAneufinder to call copy numbers for each cell.

# Clear workspace and load libraries
rm(list = ls())
library(epiAneufinder)

epiAneufinder(input="/Path/to/fragments.tsv.gz",
              outdir="./",
              blacklist="/Path/to/bed/file/containing/hg38-blacklist.v2.bed",
              windowSize = 1e5, 
              genome = "BSgenome.Hsapiens.UCSC.hg38",
              exclude = c('chrX', 'chrY','chrM'), 
              reuse.existing = TRUE,
              title_karyo="", 
              ncores = 65,
                plotKaryo=F,
              minFrags = 20000)

---

After finishing running epiAneufinder, you will get the a folder named epiAneufinder_results in the current directory. Then we should can intersect the CNV results with the cell type information to obtain the CNV profile of tumor cells.

celltype <- read.table("../1.Cellanno/HCC.celltype.txt", header = T, check.names = F)
candidate.cell <- celltype[celltype$Cell_type == "Tumor",]
candidate.cell <- candidate.cell$Index
# length(intersect(rownames(df), tumor.cell))

chrom.range <- fread("/Path/to/cytoBand.txt.gz", col.names = c("chrom","Start","End","arm","gieStain")) ## Users can download this file from zenodo
chrom.list <- paste0("chr",c(1:22, "X", "Y"))
chrom.range <- chrom.range[which(chrom.range$chrom %in% chrom.list),]
chrom.range <- dplyr::select(chrom.range, -c(gieStain))
chrom.range$arm <- substr(chrom.range$arm, 1,1)
chrom.range <- chrom.range %>%
  mutate(chrom_band = paste(chrom, arm, sep = ""))
chrom.range <- chrom.range %>%
  group_by(chrom_band) %>%
  dplyr::summarise(start = min(Start),
                   end = max(End))
chrom.range$chrom <- gsub("([[:alnum:]]+)[[:alpha:]]+$", "\\1", chrom.range$chrom_band)

# 读取calling结果
df <- read.table("/Path/to/epiAneufinder_results/results_table.tsv", row.names = 1)
input.range <- df[,1:3]
df$range <- paste0(df$seq, ":", df$start, "-", df$end)
df <- dplyr::select(df, -c(seq, start, end))
rownames(df) <- df$range
df <- dplyr::select(df, -c(range))
df <- as.data.frame(t(df))
deter.arm <- c()
for(i in 1:nrow(input.range)){
  chr.tmp <- input.range$seq[i]
  start.tmp <- input.range$start[i]
  end.tmp <- input.range$end[i]
  chrom.range.tmp <- chrom.range[which(chrom.range$chrom == chr.tmp &
                                         chrom.range$start <= start.tmp &
                                         chrom.range$end >= end.tmp),]
  deter.arm[i] <- chrom.range.tmp$chrom_band
}
deter.arm <- substr(deter.arm, nchar(deter.arm), nchar(deter.arm))
colnames(df) <- paste0(colnames(df), "(", deter.arm, ")")

rownames(df) <- gsub("cell.", "", rownames(df))
rownames(df) <- gsub("\\.", "-", rownames(df))

share.cell <- intersect(candidate.cell, rownames(df))
df <- df[share.cell,]
write.csv(df, "cancer.cnv.csv")

head(df)